Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.
Original topic: 请问在QPS没有明显变化的情况下,延迟大幅度波动可能有哪些原因?怎么调查原因?谢谢。
【TiDB Usage Environment】Production Environment
【TiDB Version】V6.1.0
【Reproduction Path】Occurs with select/update/insert
【Encountered Problem: Phenomenon and Impact】As mentioned, intermittent latency increase
【Resource Configuration】
【Attachments: Screenshots/Logs/Monitoring】
The image is not displayed. Please provide the text you need translated.
Check the traffic visualization to see if there are any particularly bright spots.
The significant change is in the p999 latency, which is usually caused by individual slow SQL queries. Slow prewrite is largely related to disk I/O and network issues. You can refer to the troubleshooting process for slow writes at TiDB 写入慢流程排查系列(一)— 前言 - TiDB 的问答社区.
At the time of the failure, there were many “key is locked” messages.
The image is not available for translation. Please provide the text content directly.
What does p99.9% mean? What does it represent? Thank you.
99.9% of the latency does not exceed the curve value, which can be compared with p99.
The image you provided is not accessible. Please provide the text you need translated.
There are several p999 spikes every day causing business alerts. Can you help provide some troubleshooting directions?
What does the following content in the TIDB log mean? What does it represent?
wait response is cancelled
Check the slow queries during the spike period in TiDB Dashboard?
Click on those highlighted bars to see which tables they are…